Wilson Ivan Guachamin Acero. Assessment of marine operations for offshore wind turbine installation with emphasis on response-based operational limits

Transcription

1 Wilson Ivan Guachamin Acero Assessment of marine operations for offshore wind turbine installation with emphasis on response-based operational limits Thesis for the degree of philosophiae doctor Trondheim, December 2016 Norwegian University of Science and Technology Faculty of Engineering Science and Technology Department of Marine Technology

2 2 NTNU Norwegian University of Science and Technology Thesis for the degree of philosophiae doctor Faculty of Engineering Science and Technology Department of Marine Technology c 2016 Wilson Ivan Guachamin Acero. ISBN (printed version) ISBN (electronic version) ISSN Doctoral theses at NTNU, 2016:336 Printed by NTNU-trykk

3 i Abstract This thesis addresses methodologies for assessment of response-based operational limits in terms of allowable vessel responses or sea state parameters, and numerical analyses of current and novel offshore wind turbine (OWT) installation activities. A generic and systematic approach that allows for identification of critical events, limiting (response) parameters and assessment of response-based operational limits of marine operations was developed. The approach is based on numerical simulations of the actual operations. Frequency and time domain techniques or operation-specific numerical analysis methods may be applied. An efficient method for assessment of the operational limits of mating operations has been proposed. The method relies on a number of crossings that a mating pin is allowed to perform out of a circular boundary (docking cone circular perimeter) in a given period of time. By carrying out a quantitative assessment of the system dynamic responses, the critical events and corresponding limiting parameters are identified. A characteristic value of a limiting parameter needs to be determined based on extreme value distributions for a target exceedance probability. This depends on the type of operation and consequences of failure events. The characteristic value is compared with the allowable limit of the limiting parameter, so the environmental conditions can be identified. The limits of these environmental parameters represent the operational limits of an installation activity. In this thesis only the effect of wave actions are considered, while current and wind actions are not included. The OWT installation activities studied in this thesis are executed with a floating heavy lift vessel (HLV). Three main activities were considered: the monopile (MP) initial hammering process, the transition piece (TP) mating operation, and the entire installation of the tower and rotor nacelle assembly (RNA). For the installation of the MP and TP, standard operational procedures were employed, while for the tower and RNA, a novel and efficient single lift installation concept was developed. This novel concept is based on the principle of the inverted pendulum and requires a cargo barge, a medium-size HLV and a specially designed upending frame. The need for huge (in terms of lifting height and capacity) offshore crane vessels is eliminated. Moreover, it has been shown that the novel concept is a feasible and valid alternative for current installation procedures. The current procedures employ jack-up vessels for sequential installation of wind turbine components; however, these activities are not studied in this thesis. The generic approach was applied to the MP, TP and tower and RNA

4 ii installation activities. For the MP initial hammering process, it was found that the critical events are failure of the hydraulic system and out-oftolerance inclination of the MP. For the TP mating operation, the critical events are the failure of the bracket supports and the failed mating attempt between the TP bottom tip and the MP. During the tower and RNA installation, structural failure of the hoist wire, structural damage of the hinged supports, and a failed mating attempt of the upending frame are identified to be critical events. For various limiting parameters, the operational limits were established in terms of allowable limits of sea states, which are a basis for assessment of the operatiblity. A methodology for weather window analysis and assessment of operability of marine operations was also developed. This methodology includes the response-based operational limits and accounts for sequence, continuity and duration of the activities, which are shown to be important in the analyses. The operational limits of the MP and TP installation were used for weather window analysis and assessment of the operability during the planning phase. Moreover, it is shown that weather forecasts can be used to identify workable weather windows and support on-board decision making during the execution phase. The methodologies provided in this thesis are systematic and efficient for modeling of current and novel OWT installation activities with the aim of establishing response-based operational limits. These are necessary for planning and safe execution of OWT installation activities.

5 iii Preface This thesis is submitted to the Norwegian University of Science and Technology (NTNU) for partial fulfillment of the requirements for the degree of philosophiae doctor. This work has been performed at the Department of Marine Technology from the Norwegian University of Science and Technology in Trondheim, with Professor Torgeir Moan as the main supervisor and Professor Zhen Gao as co-supervisor. The thesis was financially supported by the Research Council of Norway through the Centre for Ships and Ocean Structures (CeSOS) and the Centre of Autonomous Marine Operations and Systems (AMOS). This support is greatly appreciated.

6

7 v Acknowledgement I would like to express my sincere gratitude to my supervisor Prof. Torgeir Moan and my co-supervisor Prof. Zhen Gao for giving me the opportunity to carry out my Ph.D. research work in the field of marine operations. I appreciate your guidance and support for the development of my work. It has been nice working with you and a great experience in my academic life. Thanks to Prof. Günther Clauss, Prof. Ove Tobias Gudmestad and Prof. Svein Sævik, who reviewed this thesis and agreed to be part of the assessment committee. Thanks to Dr. Lin Li for her friendship and cooperation. It has been my pleasure to work with you. I also thank Professor Erin Bachynski for her friendship and valuable comments on my papers and thesis. I have very good memories of all running events with Drazen Polic and you. My good friends: Ming Song, Amir R. Nejad, Ling Wan, Xiaopeng Wu, Zhengshun Cheng, Wei Shi, Puyang Zhang, Yuna Zhao, Loup Suja, Ping Fu, Nianxin Ren, José Gallardo, Lene Eliassen, Isar Ghamari, Mahdi Ghane, Chenyu Luan, Wei Chai and Zhiyu Jiang. Thank you all for making me feel home during my staying in Trondheim. I also appreciate the support I received from other colleagues, especially from Sigrid Bakken Wold. Last but not least, I want to express my gratitude to God and my family. To my parents José and Elvia, my sister Carola and my brother Victorino for their love and support, and for being my source of motivation. To all my relatives who were always supportive when we needed help from each other, especially to my comadre Lidia, who has been my other source of inspiration. Wilson Ivan December 2016 Trondheim, Norway

8

9 vii List of Appended Papers This thesis consists of an introductory part and a summary of the papers. Methodology for assessment of operational limits Paper 1: Methodology for assessment of the operational limits and operability of marine operations. Authors: Wilson Guachamin Acero, Lin Li, Zhen Gao, Torgeir Moan Published in Ocean Engineering, Vol. 125, October 2016 OWT monopile installation Paper 2: Assessment of Allowable Sea States During Installation of Offshore Wind Turbine Monopiles With Shallow Penetrations in the Seabed. Authors: Lin Li, Wilson Guachamin Acero, Zhen Gao, Torgeir Moan Published in Journal of Offshore Mechanics and Arctic Engineering, Vol.138, August 2016 OWT transition piece installation Paper 3: Steady state motion analysis of an offshore wind turbine transition piece during installation based on outcrossing of the motion limit state Authors: Wilson Guachamin Acero, Torgeir Moan, Zhen Gao Published in Proceedings of the ASME th International Conference on Ocean, Offshore and Arctic Engineering, St. John s, Newfoundland, Canada, May 31 - June 5, 2015 Paper 4: Methodology for assessment of the allowable sea states during installation of an offshore wind turbine transition piece structure onto a monopile foundation Authors: Wilson Guachamin Acero, Zhen Gao, Torgeir Moan Under review in Journal of Offshore Mechanics and Arctic Engineering, 2016 OWT tower and RNA installation Paper 5: Numerical Study of a Novel Procedure for Installing the Tower and Rotor Nacelle Assembly of Offshore Wind Turbines based on the Inverted Pendu-

10 viii lum Principle Authors: Wilson Guachamin Acero, Zhen Gao, Torgeir Moan Under review in Journal of Marine Science and Application, 2016 Paper 6: Assessment of the dynamic responses and allowable sea states for a novel offshore wind turbine tower and rotor nacelle assembly installation concept based on the inverted pendulum principle Authors: Wilson Guachamin Acero, Zhen Gao, Torgeir Moan Published in Energy Procedia, Vol. 94, September 2016

11 ix Declaration of Authorship This thesis is composed of six papers. I have been the first author in papers 1, 3, 4,5 and 6, in which I developed the ideas, performed the analyses and reported the results. Dr. Lin Li is co-author in paper 1. She has contributed with constructive ideas for systematic development of the methodologies and provided numerical analysis results for monopile installation. She is also the first author in paper 2, in which I assisted on the development of a procedure for numerical analysis and assessment of operational limits, and provided practical input based on current industry practice. In all papers, Professor Torgeir Moan and Professor Zhen Gao have contributed with suggestions, comments and revisions that increased the scientific quality of the articles.

12 x

13 Abbreviations ν + Φ σ A B b b cr E F ext F shear h K k M Upcrossing rate Gaussian cumulative distribution function Standard deviation Added mass matrix Damping matrix Damping coefficient Critical damping Young Modulus External force vector Soil-monopile base shear force Retardation function Stiffness matrix Elastic spring coefficient Structural mass matrix m, µ Mean value r S c S allow T R Mating gap or docking cone radius Characteristic response value Allowable limit including a safety factor Reference period or duration of an activity xi

14 xii v Velocity x, ẋ, ẍ Motion, velocity and acceleration vectors CDF DAF FD FEM GBF Hs MBL MP OWT P-y PDF Q-z RAO RNA T-z TD TP Tp VMO WOWW Cumulative distribution function Dynamic amplification factor Frequency domain Finite Element Modeling Gravity based foundation Significant wave height Minimum breaking load Monopile Offshore wind turbine Soil-monopile lateral force-displacement curve Probability density function Soil-monopile end tip force-displacement curve Response Amplitude Operator Rotor Nacelle Assembly Soil-monopile axial friction force-displacement curve Time domain Transition piece Wave spectrum peak period Veritas Marine Operations Workable weather window

15 xiii Glossary of terms The definitions of the terms provided in this thesis are necessary for analysis of marine operations. Activity sequence and continuity Offshore installation activities are sequential, which means that some activities cannot start if any one of the preceding activities is not finished. Additionally, some activities are continuous and cannot be interrupted if bad weather approaches or motion responses are beyond acceptable limits. This is because in this load condition the system structural integrity may be compromised and the operation can be irreversible. Allowable limits Allowable limits are the maximum values that response parameters limiting the operations may reach to remain within acceptable safety margins. The allowable limits need to include safety factors to account for various sources of uncertainty. For the sea state parameters, the allowable limits are known as allowable limits of sea states. Similarly, for the vessel responses that can be measured on-board, for instance, motions, velocities and accelerations, the allowable limits are known as allowable limits of responses. Allowable responses Allowable responses refer to a response of a vessel in a monitoring loading condition (e.g., crane tip motion or acceleration) prior to execution which is equal to or less than the allowable limits of the responses. Allowable sea states Allowable sea states are all Hs (for corresponding T p) with values less than or equal to the allowable limits of sea states. For a group of sequential and continuous activities, the combined lower envelope will provide the allowable limits of the sea states. Characteristic value of a limiting parameter According to design codes, see e.g., Det Norske Veritas (2013a), the load effects can be represented by a characteristic value as far as possible derived from statistical data for a specified target percentile. The percentile is selected based on the type of operation and the risks associated with failure events.

16 xiv Critical and restrictive events A critical event is an occurrence that could endanger the integrity of an infrastructure, risk lives or pollute the environment, for instance the structural failure of a crane. This event is normally irreversible. On the other hand, a restrictive event does not lead to catastrophic consequences and could be reversible. For example, a failed attempt of a mating operation can be tried again, and is reversible. Governing limiting parameters This term refers to one or more parameters limiting the entire operation, i.e. resulting in the lowest allowable limits of sea states. Governing activities From a sequence of continuous offshore activities, the governing activities have the lowest allowable limits of the sea states. Limiting (response) parameters They are parameters that allow the quantification of a critical event and limit the operations. If the characteristic value of a limiting parameter exceeds its allowable limit, the safety margins are reduced and failure may occur. Marine operations According to Det Norske Veritas (2011a), marine operations are non-routine operations of limited duration to handle objects and vessels in the marine environment during temporary phases. A marine operation is a process involving interaction among the dynamic systems, operational procedures, environmental actions and human intervention. Methodology for assessment of operational limits This term refers to a sequential set of steps that are required for identification of limiting parameters and derivation of operational limits. Monitoring phase prior to execution of marine operations This phase refers to a loading condition prior to execution of an offshore activity, in which the motions of the vessels can be monitored. Notice that monitoring motion responses in this phase is different from monitoring a limiting parameter such as the wire tension during the execution phase.

17 xv Numerical methods and numerical analysis methodologies Numerical methods are frequency and time domain techniques used to find approximate solutions for equations of motion of dynamic coupled models. Numerical analysis methodologies refer to procedures developed for accurate and efficient numerical modeling of specific offshore installation activities. Operability of marine operations Operability refers to the available time for execution of a marine operation during a reference period that normally is given in terms of months or seasons. Operational limits They are allowable limits of sea states and allowable limits of vessel responses for the monitoring loading conditions prior to execution. Operational procedures The operational procedures are sets of systematic actions that provide information on the activities, sequence, duration and required sub-operations. Workable weather windows (WOWW) These are sets of continuous allowable sea states with a duration longer than the minimum required to complete a marine operation.

18 xvi

19 Contents List of Tables List of Figures xxi xxiii 1 Introduction Background and motivation Standards and guidelines for analysis of offshore wind turbine installation activities State-of-the-art OWT installation procedures Installation of foundations Installation of turbine tower, nacelle and rotor Installation of substations Novel OWT installation concepts and specialized vessels Methods for numerical analysis Operational limits Operability and weather window analysis Aim and scope Thesis outline Methodology for assessment of operational limits and operability of marine operations General Marine operation execution phases and loading conditions Assessment of the operational limits of marine operations Operational limits of a single activity Operational limits for groups of continuous activities Weather window analysis and operability Assessment of characteristic values of limiting parameters.. 23 xvii

20 xviii Contents 3 Installation systems and procedures General Monopile initial hammering process System components Installation procedure for MP initial hammering process TP mating operation using a floating crane vessel System components Installation procedure for the TP mating operation Novel concept for OWT tower and RNA installation System components Installation procedure Numerical modeling of offshore wind turbine installation activities General Numerical methods for analysis of marine operations Time and frequency domain methods Dynamic systems and numerical methods for analysis of OWT installation activities Methodologies for analysis of lift-off and mating operations Lift-off and landing operations using floating heavy lift vessels Mating operations Numerical modeling of the MP initial hammering process Numerical modeling of the TP mating operation Numerical modeling of the OWT tower and RNA installation activities Assessment of the operational limits General MP initial hammering process Identification of potential critical events and limiting parameters Assessment of dynamic responses Assessment of the MP inclination correction force Assessment of the allowable sea states for the MP initial hammering process TP mating operation Identification of potential critical events and limiting parameters

21 Contents xix Assessment of dynamic responses for the TP mating activities Motion monitoring phase prior to mating Typical installation sea states for numerical analysis of TP mating activities Axial impact between the finger guides and the MP tip Lateral impact between the finger guides and the MP outer wall Lowering and landing of the TP structure Assessment of the allowable sea states for the TP mating operation Tower and RNA installation Identification of potential critical events and limiting parameters Assessment of dynamic responses for the OWT tower and RNA installation activities Lift-off the OWT tower and RNA assembly Mating between the upending frame bottom pins and the foundation supports Upending of the OWT tower and RNA assembly Assessment of the allowable sea states for the OWT tower and RNA installation Response statistics and sensitivity study on key modeling parameters Snap forces during lift-off Impact forces and velocities during the mating activity OWT bending moment and roll during the final upending stage Operability analysis General Allowable limits of sea states for MP and TP installation Weather window analysis Operability for MP and TP installation Conclusions and recommendations for future work Conclusions Original contributions

22 xx Contents 7.3 Limitations and recommendations for future work References 89 A Appended papers 97 A.1 Paper A.2 Paper A.3 Paper A.4 Paper A.5 Paper A.6 Paper B List of previous PhD theses at Dept. of Marine Tech. 227

23 List of Tables 3.1 Main particulars of the structures for the MP initial hammering process Main particulars of the structures for the TP mating operation Main particulars of the structures for OWT and RNA installation Dynamic systems, stochastic dynamic responses and numerical methods for analysis of typical OWT installation activities. Mean value (µ) and standard deviation (σ) are statistical properties of dynamic responses Numerical methods and modeling parameters for the TP mating activities Summary of spring and damper coefficients for numerical analyses of the TP installation activities Activities, numerical methods and modeling parameters for the OWT tower and RNA installation Summary of spring and damper coefficients for the numerical analyses of the OWT and RNA installation activities Potential critical events, limiting parameters and allowable limits for the MP initial hammering process Potential critical events, limiting parameters and allowable limits for the mating operation of a TP onto a MP foundation Typical sea states for analysis of subsequent mating activities Potential critical installation activities and corresponding events considered for TD simulations Installation activity groups for weather window analysis xxi

24 xxii List of Tables

25 List of Figures 1.1 Global wind energy installed capacity Offshore wind turbine foundations and main turbine components Installation of OWT foundations Installation of turbine tower components Installation of substations Scope of the thesis and its interconnection with the papers Phases and loading conditions for execution of a marine operation General methodology to establish the operational limits Allowable limits of sea states for groups of continuous activities Weather windows analysis including continuity and duration of offshore activities. (a) Hindcast or forecasted Hs and Tp; (b) Allowable limits of Hs for corresponding Tp; (c) Hindcast or forecasted and allowable limits of Hs; (d) Dynamic responses based on forecasted wave data and allowable limits of responses for the monitoring phase; (e) Workable weather windows System components for the MP initial hammering process System components for the TP mating operation System components during lift-off of the OWT tower and RNA Installation procedure for OWT tower and RNA. (a,b) Motion monitoring phase; (c) Mating between the upending frame and the foundation supports; (d) Upending of the OWT tower in the final stage; (e) Completion and upending frame removal xxiii

26 xxiv List of Figures 4.1 Determination of impact and snap velocities based on TP bottom tip displacement and velocity TD histories. (a) TP bottom tip relative heave displacements (respect to the crane tip) for constant lowering speed; (b) Possible impact and snap velocities (no winch speed included) Spring-damper models. (a) Gripper-MP physical and contact models; (b) Soil-MP interaction model Typical soil reaction moment vs MP inclination due to cyclic loading with a period of 6 s Schematic outline of a dynamic coupled model for the MP initial hammering process Schematic outline of a dynamic coupled model for TP landing Models for TP-MP axial contact-impact prior to mating. (a) Physical model; (b) Contact points; (c) Spring-damper model (for Activity 2) Lateral contact during TP mating operation. (a) Physical model; (b) Spring-damper model (for Activities 2, 3) Contact-impact during TP landing (a) Physical model; (b) Spring-damper model (for Activity 4) Schematic outline of a dynamic coupled model for the tower lift-off and upending frame mating operation Elastic contact models. (a) OWT tower-cargo barge interaction model for lift-off and mating (Activities 1, 2 and 3); (b) Spring-damper model for the mating phase (Activity 3); (c) Tower upending articulation model (Activity 4) Standard deviation of MP inclinations and contact forces on one hydraulic cylinder at different MP penetration depths (pene) and wave condition from 3 hour time domain simulations, Hs= 1.5 m, dir= 150 deg Correction force vs maximum inclination (over 10 min) of the MP for various penetration depths (pene) Allowable limits and extreme responses, Hs= 1.5 m, wave dir= 150 deg. (a) MP-gripper contact forces, exceedance probability 10 4 based on 3 hours TD simulation; (b) MP inclination over 10 min simulations Allowable limits of sea states for the initial phase of the MP hammering process

27 List of Figures xxv 5.5 Allowable limits of sea states for the monitoring phase of the TP s bottom tip. Limiting parameter: TP s bottom tip motions, allowable crossing rate limit: ν + allow = Hz and mating gap: r = 0.3 m Response statistics of the axial impact velocities prior to the mating phase. Based on the collision approach method Example of dynamic responses from TD simulations of the lowering and landing phases. Hs = 1.60 m, T p = 6 s, α = 135 deg Allowable limits of sea states for TP-MP mating. Assumed allowable limit for the motion monitoring phase: crossing rate ν + allow = Hz for r = (0.3, 0.5) m, arbitrary allowable limits for the landing phase: v imp = (0.10, 0.18) m/s Hoist wire tension during tower lift-off. (a) Hs=1 m, Tp=8 s; (b) Hs=1 m, Tp=6 s Impact forces on the foundation supports during the mating phase. (a) Hs=1 m, Tp=8 s; (b) Hs=1 m, Tp=6 s Dynamic responses during the OWT upending operation. PM wave spectrum Hs=1 m, Tp=8 s Allowable limits of sea states for the lift-off and mating operations, allowable limits: F snap = 5000&7000 kn (Number of seeds= 60, simulations duration 1 min), pin motion (ν + = Hz, r = 0.35 m) Statistical parameters of the snap forces during OWT tower lift-off. (a) Sensitivity study on seed numbers, wave dir= 120 deg; (b) Snap force statistical parameters for the allowable limits of sea states given in Fig for various wave dir., No. seeds= 60. Duration of TD simulations for the snap load events 1 min Maximum snap forces during OWT tower lift-off, wave dir=160 deg, No. seeds= 60. (a) Sensitivity to hoist wire stiffness; (b) Sensitivity on winch lifting speed. Duration of TD simulations for the snap load events 1 min Sensitivity study on contact spring stiffness during the mating operation, wave dir=160 deg, No. seeds= 60. (a) Maximum impact forces; (b) Maximum impact velocities. Number of seeds= 60, duration of TD simulation for impact events 1 min

28 xxvi List of Figures 5.16 Sensitivity study on rotational spring coefficient k r during the final stage of the OWT upending operation, Tp= 7 s, various headings, No. seeds= 60. (a) Maximum articulation reaction moments; (b) Maximum out of plane OWT inclination. Duration of each simulation 15 min Allowable limits of sea states for single activities and activity groups (see. to Table 6.1) for various headings. (a) Allowable limits of sea states for the MP lowering operation; (b) Allowable limits of sea states for the MP initial hammering process; (c) Allowable limits of sea states for group G2; (d) Allowable limits of sea states for group G Typical weather window analysis based on hindcast wave data at universal time coordinate (UTC), and heading into the waves. a) Hindcast and allowable limits of Hs (for corresponding Tp); b) Workable weather windows Operability for MP and TP installation for various headings and months

29 Chapter 1 Introduction 1.1 Background and motivation The need for alternative energy sources others than fossil fuels has recently increased. The depletion of oil reservoirs, environmental pollution and global warming are leading into a transition from fossil fuels into renewable alternatives such as wind energy. In recent years, wind energy has grown fast due to its potential and increasing demand. Figure 1.1 shows that by the end of 2015 the cumulative installed capacity was approximately 432 GW. This amount includes GW installed offshore, from which GW were added in The European Wind Energy Association (2015b) has forecasted that by 2030, the offshore wind energy installed capacity in the European Union (EU) countries can be 66.5 GW. Installed wind capacity [GW] Figure 1.1: Global wind energy installed capacity (Global Wind Energy Council, 2015) To satisfy the future energy demands, more offshore wind turbines will 1

30 2 Introduction be installed. The current tendency is to move towards offshore locations, because the wind velocity is more steady and higher, larger offshore wind turbines (OWTs) can be installed when compared with onshore wind turbines, and there is less impact on people, e.g., noise. Offshore wind turbines are structures installed in open waters with the aim of generating electricity from the energy in the wind. Depending on the water depth, several concepts for OWT foundations exist. Figure 1.2 shows that the main components of an OWT are the foundation, the turbine tower, the nacelle and the blades. The foundation is composed of the bottomfixed or floating substructure and a transition piece (TP) connecting the foundation and the turbine tower. Bottom-fixed foundations such as tripods and jackets are fastened to the seabed by means of piles. Floating concepts are connected to the seafloor using mooring lines and tendons. By the end of 2015, 3230 OWTs with bottom-fixed foundations were grid-connected in the EU countries. From these foundations, 80 % were monopiles (MPs), 9.1 % gravity-based, 5.4 % jackets, 3.6 % tripods and 1.7 % were tripiles (European Wind Energy Association, 2015a). The reason for the preference of MPs over other foundations is because of their construction and installation simplicity, and lower associated costs. As the water depth, size, and number of the wind turbines increase, the trend is to use larger diameter monopiles and shift from MPs to jackets and tripods, see Fig According to the European Wind Energy Association (2015a), during 2015, 419 offshore wind turbines with an average capacity of 4.2 MW were installed in an average water depth of 27.1 m. Furthermore, it is expected that the average capacity of new wind turbines will increase to 8 MW by During the life cycle of an OWT, the installation phase is important. This is because some structural components can experience the largest loads in their lifetime. In addition, the cost of the installation phase is significant, and would correspond to approximately 20% of the total capital expenditure (CAPEX) (Moné et al., 2013). Since the profit margins in the wind energy sector are low, it is necessary to reduce these costs by improving the current installation procedures, developing more efficient installation concepts, using more accurate and systematic methodologies for analysis of marine operations, and increasing the operational limits of the installation activities while maintaining sufficient safety margins.

31 1.2. Standards and guidelines for analysis of offshore wind turbine installation activities 3 blades nacelle tower TP GBF Monopile Tripod Jacket TLP Semi-sub Spar Figure 1.2: Offshore wind turbine foundations and main turbine components 1.2 Standards and guidelines for analysis of offshore wind turbine installation activities Planning and execution of OWT installation activities are carried out according to recommendations provided by regulatory authorities. To date, the modeling and execution of OWT installation activities are mainly based on guidelines that have been developed for the oil and gas offshore industry. Recommended practices and guidelines such as International Organization for Standardization (2015a) and Det Norske Veritas (2010) are suitable for modeling environmental conditions and loads. Some recommendations and simplified formulations for analyses of typical lifting operations are provided by Det Norske Veritas (2011b). Based on these analyses, the response parameters of the dynamic systems can be studied and used to establish operational limits. These operational limits and other relevant parameters such as duration of the operations, forecast lead time, etc., are required for planning of marine operations. General requirements for these parameters are provided by Det Norske Veritas (2011a), GL Noble Denton (2015), London Offshore Consultants Limited (1997) and International Organization for Standardization (2009, 2015b). Among them, the last one provides guidelines and requirements for modeling and analysis of marine operations for offshore wind turbine installation during planning phases, and operational aspects for execution phases. An important parameter for analysis of marine operations is the duration

32 4 Introduction of the operation. This duration is referred to as reference period T R by Det Norske Veritas (2011a), and is defined as: T R = T P OP + T C where T P OP is a planned operational period and T C is a contingency time. If the reference period is less than 72 hours, the marine operation is weather-restricted, and weather forecasts can be used for decision-making prior to execution of marine operations. The focus of this thesis is on offshore wind turbine MP, TP, and tower and RNA installation activities, which are normally weather-restricted. 1.3 State-of-the-art OWT installation procedures This section describes the most commonly used and novel procedures for installing offshore wind turbine components and assemblies of these components Installation of foundations Installation procedures for OWT bottom-fixed foundations are similar to the ones used in the offshore oil and gas industry. Gravity based foundations (GBFs) are suitable for shallow water. The mass can be over 2500 tons (Thomsen, 2014), and thus, large heavy lift crane vessels are needed for a single lift operation. However, the normal procedure is to tow them to the site, and ballast and lower them in a controlled sequence onto the specially prepared seabed. The most common type of bottom-fixed foundation is the monopile, which can be employed in water depths up to approximately 40 m (Thomsen, 2014). The MP is normally transported on the deck of the HLV which can be a floating platform or a jack-up barge. It is also common to transport the MPs in dry condition using feeder cargo barges. Once at the offshore site, the crane lifts-off, upends, and lowers the MP to the seabed. Then, a hydraulic hammer is used to drive the MP to final penetration. Tripods and jacket structures are normally transported on the deck of the installation vessels or transportation barges. The foundations are liftedoff, positioned, and lowered to the seabed using an on-board crane. A hydraulic hammer is used to drive small diameter piles and fasten the foundations to the seabed. Then, a grouting mixture is poured in the annular gap between the piles and foundation leg sleeves. Because of the size of the tripod and tower structures, large HLVs in terms of height and lifting capacity are required, see Fig. 1.3.

33 1.3. State-of-the-art OWT installation procedures 5 In addition, a transition piece (TP), which connects the MP and sometimes the tripod foundations with the turbine tower, needs to be installed. The TP is normally lifted-off from the own deck of the installation vessel, aligned, lowered onto the MP foundation, leveled and finally grouted. Leveling the TP is required to adjust the final inclination of the OWT tower. (a) Tripod installation using a jack-up vessel. Source: (b) Jacket installation using a floating HLV. Source: com Figure 1.3: Installation of OWT foundations Installation of turbine tower, nacelle and rotor The installation of the turbine tower components is new compared to the oil and gas industry practices. Nowadays, the split installation procedure as shown by Wang and Bai (2010), is commonly used to install the tower, nacelle and each blade separately and in sequence. The operations are executed with jack-up crane vessels, see e.g., Fig. 1.4 (a). There are also variations of this procedure with the aim of reducing the number of lifts, and thus, the installation time. The drawback of the split method is that it is weather-sensitive during the standard jack-up leg lowering and retrieval process. The allowable limit for the significant wave height (Hs) is between 1.2 and 1.5 m (Thomsen, 2014). Moreover, the operational water depths are between 30 to 60 m (El-Reedy, 2012). Alternatively, a fully assembled tower and RNA has been installed by a single lift procedure using a huge floating crane vessel, see Fig. 1.4 (b). This operation was found to be very challenging and unattractive (Sarkar, 2013). A numerical analysis of this operation was performed by Ku and Roh (2015) by applying time domain simulations.

34 6 Introduction (a) OWT balde installation using a jack-up vessel. Source: licdn.com/mpr/mpr/shrinknp_800_ 800/p/6/005/0b7/104/353401b.jpg (b) Single lift OWT and RNA installation using a floating HLV. Source: rambiz-3000/31/ Figure 1.4: Installation of turbine tower components Installation of substations In addition to the foundations and turbine towers, another important wind farm component is the substation. It is a cluster infrastructure designed to collect the energy generated by every wind turbine and convert it into high voltage current before transporting it to shore using power transmission cables. The installation of this module is normally carried out by the single lift or the float-over method, see Fig The choice of the method depends on the available capacity of the HLVs and the mass of the module. The single lift method is normally used when the module is not too heavy for the available HLVs, see Fig. 1.5 (a). The installation procedure consists of the following steps: lifting the module from the deck of the HLV or from a transportation barge, positioning it a few meters above the foundation, aligning the mating pins, monitoring the motion responses of the pins, lowering, mating and landing. For topsides with large mass and size, the float-over method is commonly applied, see Fig. 1.5 (b). In this method, a cargo barge that transports the module in dry condition, floats over the specially designed jacket. After that, the barge is moored to the jacket, ballasted, and the mating phase occurs. Next, the barge is ballasted further to achieve enough clearance between the structures, and finally the barge sails away. It is observed that the installation of a substation and a transition piece (see Subsec ) share some similarities in the execution procedure. There is a phase where the mating points are aligned and the motions are moni-

35 1.3. State-of-the-art OWT installation procedures 7 tored. If the motions are too large when compared to the allowable mating gaps, the subsequent operations may not be possible. This is an important criterion for making a decision on-board vessels during execution of installation activities. (a) Substation installation using an HLV. Source: (b) Substation installation using the float-over method. Source: offshore-wind/hvdc_sylwin_alpha Figure 1.5: Installation of substations Novel OWT installation concepts and specialized vessels New installation vessels and concepts have also been proposed. Huisman Equipment B.V. (2015) presented a concept of a new twin hull installation vessel which uses an active heave compensation system during landing of the tower on the foundation. Seok (2013) proposed a method to install an OWT tower using a specially modified jack-up vessel. This vessel has an opening that allows the OWT foundation to get into it. This is done by varying the tension and length of the mooring lines followed by a standard leg lowering procedure. A frame on the deck allows for vertical positioning and lowering of the OWT tower. However, this installation vessel and procedure have not been applied. Sarkar and Gudmestad (2013) proposed a novel concept for the installation of an OWT tower and RNA. The procedure requires a patented system for keeping the blades out of the water during the tower transportation. The technical feasibility of this concept was based on some preliminary design considerations and operational aspects carried out by Korovkin (2012); however, the method has not been applied in practice. A self installing concept of a fully integrated tower and gravity based foundation was proposed by Wåsjø et al. (2013). In this concept, the assembly is

36 8 Introduction transported on two cargo barges and is installed using a controlled lowering process. Model tests and numerical analyses have been conducted to assess its feasibility (Bense, 2014). Furthermore, Jin and Jo (2014) suggested the use of a stabilizing frame to constrain the pendulum motions of a fully assembled tower that is lifted from a cargo barge. Finally, Graham (2010) developed a concept that consists of a cargo barge transporting the OWT in horizontal position to the site and a mechanism to rotate the tower to the vertical position. The rotation is carried out using hydraulic systems located on the deck of the cargo barge. After that, pins located at the bottom of the tower mate with the docking cones of a specially designed foundation. Next, the stern of the barge is pulled backwards using winches and anchors on the seabed. Then, a clamping system on the foundation receives the tower in a vertical position. However, this method has not been shown to be feasible and has not been applied in practice. Based on the above-given literature review, it appears that there are not many alternative concepts for OWT tower installation; the proposed novel solutions require new vessels and procedures for which very limited information exists regarding numerical analyses and feasibility. Thus, alternative installation methods which are feasible, efficient, and use available equipment are required. 1.4 Methods for numerical analysis Analysis of marine operations requires a synergy between various disciplines such as structural mechanics, hydrodynamics, stochastic methods and probabilistic design. This is necessary to develop numerical modeling procedures and computer codes that allow for representing the actual marine operations. In addition, the theory implemented in each computer code varies depending on the application, and in general a combination of various computer codes is necessary to numerically model marine operations. Computer codes such as OFFPIPE (Malahy, 2013) and OrcaFlex (Orcina Ltd, 2006) are commonly used for analysis of pipeline, cable and umbilical installation. These codes can provide reaction forces and stresses on the pipeline; however the motion response amplitude operators (RAOs) of the installation vessel are normally computed in other codes such as Wamit (Lee, 1995), Wadam (Det Norske Veritas, 2013b) or AQWA-LINE (Century Dynamics-Ansys Inc., 2011). The computer code SACS (Structural Analysis and Design Software) has modules to analyze structural responses during offshore transportation and jacket launching, but the RAOs need to be in-

37 1.4. Methods for numerical analysis 9 put as well. For offshore heavy lifting operations, there are computer codes that perform well with motion response analysis but offer limited number of mechanical couplings and structural analysis capabilities. Some examples are: OrcaFlex, the Ansys-AQWA suite of programs, SIMO and Riflex (MARINTEK, 2012) and anysim (MARIN, 2009). The dynamic responses obtained from these codes can serve as input into other more specialized structural or fluid dynamics codes. Thus, a number of computer codes are necessary to build even simplified numerical models and assess the dynamic responses of structures during planning of marine operations. Dynamic responses need to be assessed by numerical modeling of the actual marine operation. Depending of the type of marine operations, this can be done by applying frequency domain (FD) and time domain (TD) methods. In general, TD are applied to study dynamic responses of non-linear systems and systems with time-variant dynamic properties. Taking as an example the float-over operation, the dynamic system needs to include non-linear contact elements, because contact between mating structural elements will occur during the entire mating operation. Moreover, the dynamic properties such as natural periods of the system may change during the installation, because the cargo barge is ballasted. Thus, the mating process is non-stationary, and as a result of this process, the mean vertical position of the barge changes with time. Due to the complexity that involves modeling the actual float-over operation and the limited number of features of existing computer codes, the current practice is to conduct TD simulations of float-over operations with the barge at specific draughts or percentages of load transfer (Jung et al., 2009; He et al., 2011). Thus, the actual non-stationary float-over operations are not simulated, so the dynamic responses can be expected to be different from cases when the actual operations are modeled numerically. Non-stationary lifting and landing operations have been studied by Sandvik (2012). It was found that the response statistics resulting from numerical analyses of non-stationary processes are significantly different from cases where operations were modeled using simplified stationary process TD simulations. This is because in a non-stationary lifting or lowering process, the dynamic properties of the system change with time, the dynamic responses do not build-up as in a stationary case, and in the stationary case the actual duration of the operation is not modeled. Thus, numerical modeling of actual operations is necessary. As it was shown above, modeling marine operations is challenging, and numerical models are only approximate representations of actual dynamic systems. In addition, important features cannot be captured with existing

38 10 Introduction modeling techniques and computational tools. Thus, more accurate numerical analysis methodologies are needed. Li et al. (2014b, 2015b) developed a methodology to account for shielding effects of installation crane vessels on monopile foundations and the radiation damping of the monopile during non-stationary lowering process. In addition, efficient numerical analysis methodologies are required for quick assessment of the dynamic responses. Fylling (1994) proposed a methodology that is suitable for estimation of initial impact velocities of drifting vessels. This methodology was applied by Sandvik (2012) to study lift-off and landing operations of dynamic systems with time-invariant dynamic properties. This method is suitable for assessment of snap and impact velocities by imposing a winch speed on time histories of displacements obtained from stationary process TD simulations. Moreover, a simplified method to assess snatch loads during OWT monopile lift-off has been proposed by Zhao et al. (2016). Based on the literature review, few efficient and accurate methodologies for numerical modeling of marine operations have been developed. In addition, these methodologies are operation-specific and are not applicable for other marine operations. Moreover, the features and tools available in the state-of-the-art codes are not sufficient to model entire marine operations. Instead, the operation is divided into several activities and some of them may not correspond to the actual operations. Thus, further development of methodologies and tools is required. 1.5 Operational limits During execution of a marine operation, the basic safety criterion is that any response of the dynamic system (e.g., the wire tension) should not exceed its allowable limit (including safety factors). However it is practical to transform these limits into limits in terms of sea state parameters or motion responses of vessels that can be monitored before and during the execution of marine operations. These are operational limits that can be established during the planning phase and should be derived systematically. Moreover, these limits need to account for uncertainties in material mechanical properties, statistical uncertainty, model uncertainty, forecast uncertainty, etc. Depending on the type of operation and consequences of failure events, safety factors need to be established and calibrated to ensure enough safety margins. Det Norske Veritas (2011a), International Organization for Standardization (2015b) and GL Noble Denton (2015) provide recommendations on the operational criteria for planning and execution of marine operations. The

39 1.5. Operational limits 11 criteria are mainly expressed in terms of significant wave height (Hs), while the wave spectrum peak period (T p) is not considered. The Hs parameter is reduced by alpha factors that account for uncertainties in the weather forecast methods and leading times, and the reference period (duration) T R of the activities. However, floating units are highly sensitive to T p, and thus, T p also needs to be included. Clauss and Riekert (1990a,b) presented a summary of operational limits in terms floating crane vessel motion responses. These limits were given based on experience from projects executed in the North Sea. Some of these vessel motion criteria were also expressed in terms of sea state parameters. Likewise, Smith et al. (1996) provided the operational limits in terms of allowable impact velocities for a jack-up vessel during the standard leg lowering procedure. The limits were derived from structural damage criteria based on structural analyses of leg members. Similarly, Clauss et al. (1998) proposed a methodology for assessment of the allowable sea states during offshore pipelaying based on maximum permissible stresses on the pipe. The methodology accounts for wave and vessel motion induced stresses. In addition, Cozijn et al. (2008) assessed the operational limits for installing a module using a floating crane semi-submersible platform onto a floating vessel. The limits were derived based on numerical analysis, model tests, and offshore site measurements. Moreover, Graczyk and Sandvik (2012) established the allowable sea states for the lift-off and landing of an offshore wind turbine component on the deck of a ship. The dynamic response of the lifted object was estimated based on formulations given by Det Norske Veritas (2011b) and the allowable acceleration on the lifted object was simply assumed. An approach to derive the operational limits in terms of Hs and T p for a drilling jack-up unit during the deployment and retrieval of its legs was given by Matter et al. (2005). The allowable stresses in the spud cans, legs, and pinions were established based on structural analyses. These allowable stresses were expressed in terms of allowable vessel motions, and by using the response amplitude operators (RAOs) in a free floating condition, these motions were given in terms of allowable sea states. Similarly, Ringsberg et al. (2015) presented the allowable sea states for a jack-up vessel during deployment of its legs. The sea states were identified by comparing the allowable forces on the spud can, which were derived from finite element modeling (FEM), with the characteristic values of the impact forces, which were computed from a coupled spud can and soil interaction model. The literature cited above shows that the operational limits for marine operations have been assessed considering different approaches, which vary

40 12 Introduction and are not clearly indicated. 1.6 Operability and weather window analysis The operability is a measure of the available time for executing an operation in a given reference period and in a safe manner. It is normally assessed using the operational limits (in terms of sea state parameters) and scatter diagrams of the offshore site. Human comfort criteria for operability analysis have been established in the past (NORDFORSK, 1987; Lawther and Griffin, 1987; Werenskiold et al., 1999). These criteria have been used to assess the operability of ships. For instance, Fonseca and Guedes Soares (2002) studied the operability of a container ship and a fishing vessel. Several criteria such as vessel roll and deck accelerations were considered, and a sensitivity study on the most relevant parameters was carried out. In addition, Tezdogan et al. (2014) assessed the operability of a high speed catamaran vessel based on passenger comfort criteria. A comparative study by applying various sea-keeping theories was conducted, and the effect of seasonality was also investigated. The operability of OWT service vessels has also been assessed. For instance, Wu (2014) assessed the operability for the docking operation between service vessels and OWT foundations. This was done for the current access method that relies on the friction force between the vessel and the foundation. Dynamic responses computed using the frequency domain method and scatter diagrams were employed. The operability should preferably be assessed from weather window analysis, where the sequence, duration, and continuity of each activity can be included. Nielsen (2007) provided a procedure to estimate the available time for execution of a marine operation. The procedure is based on the conditional distribution function of Hs on the duration of weather windows, so the time histories of hindcast wave data are employed. This method is suitable for planning of marine operations; however, the peak period T p is not included. Bergøe (2015) provided the operability of jack-up and floating units. Although the allowable sea states were simply assumed, the sequence and duration of the activities were incorporated in the analyses. In addition, Velema and Bokhorst (2015) identified the weather windows for installation of a subsea storage module. The heading providing the best responses was selected based on directional wave spectra from updated weather forecast and on-board numerical simulations. Moreover, Gintautas et al. (2016) proposed a methodology for identification of weather windows with the aim to support on-board decision-making during offshore wind turbine installation.

41 1.7. Aim and scope 13 This is done by on-board numerical simulation of the operations and probabilistic assessment of the dynamic responses, which are computed using updated forecast wave data. In the analyses, the sequence and duration of the activities were included; however, the operational limits were simply assumed. Based on the aforementioned literature, it appears that no methodologies that link practical and systematically derived response-based operational limits and weather window analysis have been published. These operational limits should be used to assess the operability of different OWT installation procedures while considering the duration, sequence and continuity of the activities. 1.7 Aim and scope In view of the lack of published methods for establishing transparent and practical operational criteria for marine operations, and the need for feasible alternative OWT installation procedures, important objectives of this thesis are: Establish a general and systematic methodology for identification of critical events and assessment of the operational limits for marine operations. Establish practical operational limits in terms of Hs, T p, and vessel response parameters that can be monitored on-board and used for decision-making. Develop a methodology for operability assessment of marine operations that accounts for response-based operational limits, duration, sequence and continuity of the activities. Develop procedures for numerical modeling of current OWT installation activities. Develop methodologies for numerical analysis of critical OWT installation activities, e.g. the mating operation of a transition piece. Develop alternative and feasible OWT installation concepts or procedures. This thesis is written as a paper collection, including five journal papers and one conference paper, which are attached in the Appendix. The scope

42 14 Introduction of the thesis is shown in a matrix in Fig. 1.6, where the main topics and the interconnection between appended papers are illustrated. OWT installation activities using a floating heavy lift vessel Tower and RNA installation MP installation TP installation Split install. method, etc. Development of a novel concept MP and TP mating operation Lift-off, lowering, etc. MP initial hammering process Lift-off, lowering, etc. Paper 5 (Feasibility study Inverted pendulum) Paper 3 (Crossing rate for mating operations) Paper 6 (Dyn. responses & oper. limits) Paper 4 (MP and TP mating) Paper 2 (MP initial hammering process) Installation procedure Numerical modeling Paper 1 (case study) Paper 1 (case study) ID limiting parameters Operational limits Weather windows, operability Paper 1 (Development of methodology for assessment of operational limits) Figure 1.6: Scope of the thesis and its interconnection with the papers Figure 1.6 shows the research approach that is followed in this thesis. A

43 1.8. Thesis outline 15 generic methodology for systematic derivation of operational limits for marine operations is developed (paper 1 ). The generic methodology is based on numerical modeling of actual (in general) non-stationary operations and quantitative assessment of dynamic responses. By comparing characteristic values of dynamic responses with their allowable limits, a backward derivation of limits of sea state parameters is possible. This methodology is applied to MP, TP and OWT and RNA installation activities. For the MP and TP (papers 2 and 4 ), standard installation procedures are considered, while for the OWT tower and RNA (paper 5 ), a novel installation concept is developed. Numerical models and methods are developed for the installation activities. A numerical method suitable for mating operations is addressed in paper 3. The numerical models are established based on information available in literature and assumptions. Meanwhile, the allowable limits of some response parameters are estimated based on specifications of the equipment and reasonable assumptions. Contact-impact problems are only addressed in terms of representative global motion responses, e.g. initial relative impact velocities, so the structural damage criteria which need to be established based on FEM are not included. For some installation phases, sensitivity studies on modeling parameters are carried out (paper 6 ). Only wave load actions are considered, while wind and current load actions are not included. 1.8 Thesis outline This thesis is composed of seven chapters. The content of each chapter including the reference papers is briefly summarized below. Chapter 1: This chapter provides an introduction including the background and motivation, a brief review of current guidelines and standards, an overview of current OWT installation methods, some methodologies for accurate and efficient numerical analysis of marine operations, current developments in operational limits, the aim and scope, and the outline of the thesis. Chapter 2: This chapter deals with a general methodology for the assessment of the operational limits and operability of marine operations, especially for offshore heavy lifting. This is a summary of paper 1. Chapter 3:

44 16 Introduction This chapter addresses the installation procedures and system components used for numerical modeling of the MP, TP and OWT tower and RNA installation activities. This chapter is a summary of papers 2, 4 and 5. Chapter 4: This chapter provides the numerical methods and models that were used for analysis of various OWT installation activities. The numerical methods are provided as a summary of papers 3 and 4, while the numerical models are a summary of papers 2, 4 and 5. Chapter 5: In this chapter, the operational limits for the installation activities are established. They are assessed by applying the methodology provided in Ch. 2. Numerical models given in Ch. 4 are employed. The results are a summary of papers 2, 4, 5 and 6. Chapter 6: In this chapter, the operability for installation of the MP foundation and the TP using a floating crane vessel is provided. This chapter is a summary of a case study in paper 1 Chapter 7: Conclusions, highlight of original contributions and recommendations for future work are provided.

45 Chapter 2 Methodology for assessment of operational limits and operability of marine operations 2.1 General For any marine operation, the feasibility of its execution needs to be evaluated based on the characteristics of the vessels, equipment and environmental conditions of the offshore site. If it is feasible, the equipment and installation season need to be selected well in advance. This is achieved by assessing the performance of the installation system during execution of these activities. Numerical simulations, documentation, guidelines and experience from related past projects are normally employed to establish the operational limits, which can be used for assessment of the operability during the planning phase or for decision-making during the execution phase of marine operations. The current industry practice for planning and execution of marine operations, is to express the operational limits in terms of Hs design limits. These limits are mainly based on experience from past projects. Moreover, offshore standards provide guidelines to account for forecast uncertainties in the Hs parameter. This is done for instance by means of alpha factors, which are reduction factors (applied on Hs) that depend on the forecast methods and duration of the operations (Det Norske Veritas, 2011a). However, the approach that is followed to establish these Hs design limits is not 17

46 Methodology for assessment of operational limits and operability 18 of marine operations clear, and the following questions may be asked: How should the operational limits be established? Are the Hs limits sufficient? Which installation activities should be chosen for numerical analyses? How should numerical simulations be conducted? Which response parameters should be analyzed? How should response statistics be assessed? The above questions have not been properly addressed in the literature and scientific papers. Thus, a systematic methodology for assessing the operational limits and operability of marine operations is required. Guachamin Acero et al. (2016d) (paper 1 ) developed a general approach that is suitable for identification of critical events and corresponding limiting (response) parameters of critical operations, and assessment of the operational limits in terms of allowable limits of sea states. Moreover, a general procedure for assessment of the operability of marine operations based on weather window analysis was proposed. A brief description of these procedures is provided in this chapter. 2.2 Marine operation execution phases and loading conditions Figure 2.1 shows two phases during the execution of a marine operation. First, there is a monitoring phase prior to the (actual) execution of marine operations (DYNAMIC SYSTEM 1), in which the responses of the vessel e.g., motions, velocities, accelerations are monitored and compared with the operational limits given in the operational manual. This is done to make a decision on whether or not to start an operation. In this loading condition, the system is weakly non-linear, it has time-invariant properties, and the resulting processes are stationary, see section 4.2. Thus, frequency domain (FD) methods can be applied. This is suitable for computations using on-board systems. Second, there is an execution phase with different loading conditions in which critical events can occur (DYNAMIC SYSTEM 2). These loading conditions are necessary for numerical analysis and assessment of the operational limits. Moreover, during the execution of the activities, some

47 2.3. Assessment of the operational limits of marine operations 19 DYNAMIC SYSTEM 1 Monitoring phase DYNAMIC SYSTEM 2 Execution phase Loading condition LC 1 Loading condition LC 2 Operation begins Make decisions on whether to start or not the operation based on measured vessel responses, forecasted sea states and operational limits YES Execute the offshore activities Operation ends Figure 2.1: Phases and loading conditions for execution of a marine operation (Guachamin Acero et al., 2016d) dynamic responses can be monitored, for instance, the wire tension. This parameter can be used to take mitigation actions (if the tension reaches dangerous levels), but not to make decisions before executing an offshore activity. 2.3 Assessment of the operational limits of marine operations This section deals with a methodology for assessment of the response-based operational limits of a marine operation. The operational limits are derived from numerical analysis of the actual execution phase. These limits can be used in the monitoring phase prior to execution to support on-board decision-making, and during the planning phase for operability analysis Operational limits of a single activity The methodology for assessment of the operational limits is shown in Fig Based on typical installation procedures, the first objective is to narrow down the number of cases for numerical analyses. Preliminary identification of potential critical events and their corresponding activities is commonly done by applying qualitative risk assessment techniques together with technical discussions form past projects and reports, see step 2 in Fig These activities are selected to build numerical coupled models and carry out global dynamic response analyses of the system under typical sea states (steps 3 and 4). Quantitative assessment of the dynamic responses

48 Methodology for assessment of operational limits and operability 20 of marine operations is conducted (step 5), and the parameters that may reach dangerous levels (limiting parameters) when compared with their allowable limits (including safety factors) are identified (step 6). For contact-impact problems, the allowable limits need to be established based on structural damage criteria, see e.g., Li et al. (2014a). For time domain simulations conducted in step 4, the number of seeds required for assessment of the dynamic responses does not need to be large, because these responses are only used for screening of limiting (response) parameters and are not used for assessment of the operational limits. Note that the actual operations need to be simulated numerically, i.e. the stationarity, linearity and time-variant properties of the dynamic system need to be considered. Time domain and frequency domain methods are normally applied. Moreover, more accurate and efficient methodologies for numerical analysis of specific marine operations have been developed by Li et al. (2014b, 2015b) and Sandvik (2012). Some methods used in this thesis are addressed in Sec For the identified limiting (response) parameters, the corresponding dynamic coupled models (step 7) are used to carry out numerical simulations, and build response statistics and extreme value distributions. This is done for all possible Hs and T p combinations. The characteristic value of a limiting parameter is selected, e.g., for a target probability of non-exceedance based on extreme value distributions (step 8). This probability depends on the type of operation, duration and consequences of the failure events, see Sec For the time domain simulations carried out in step 8, normally a large number of seeds is required to achieve convergence on the response statistics. In equation (2.1), the characteristic values S c (Hs, T p) are compared with the allowable limit S allow (including a safety factor γ s ) of a limiting (response) parameter, and thus, the allowable limits of sea states are established, see step 9. Finally, the allowable limits of sea states can be expressed in terms of allowable limits for responses of the system in a loading condition prior to execution, see monitoring phase in Fig These limits are useful because they can be compared with responses that are measured on-board. Figure 2.2 shows that allowable limits of both, sea states and responses in monitoring phases are known as operational limits, and they should to be provided in the operational manuals. S c (Hs, T p) = 1 γ s S allow (2.1)

49 2.3. Assessment of the operational limits of marine operations 21 - Executeactivity1 - Executeactivity2 - Executeactivity3 Reliability qualitative methods: FTA, FMEA, root cause, etc 1. Installation procedure 2. Identify potentially critical events Input typical sea states or any other environmental parameters 3. Numerical models for each potentially critical installation activity (e.g. DYN. SYSTEM 2 in Fig. 2.1 Is structural damage criteria needed? N - Loadmagnitude - Consequences Input all possible combinations of Hs and Tp 4. Global dynamic analyses 5. Assessment of dynamicresponses 6. Identify critical events and limiting parameters 7. Select critical marine operation coupled dyn. models 8. Compute characteristic values of dynamic responses (S c ) for limiting parameters Y FEM of thethe structural components Assessstructural damagecriteria Allowable limits (S allow ) 9. Allowable sea states for S c S allow or operational limits Operational limits Hs --- Lim. param1 Lim. param2 Hs Allow. limits of sea states (for governing parameter) z Allow. limits of motion responses for monitoring phases prior to execution Based on DYN. SYSTEM 2 Based on DYN. SYSTEM 1 Tp Tp Tp Figure 2.2: General methodology to establish the operational limits (Guachamin Acero et al., 2016d)

50 Methodology for assessment of operational limits and operability 22 of marine operations Operational limits for groups of continuous activities The installation of an OWT normally requires the execution of several sequential activities. These activities can be continuous, and thus, they can be grouped, e.g., MP lift-off and lowering. By following the procedure given in the previous subsection, the allowable limits of sea states for each activity can be established. Figure 2.3 shows two groups of continuous activities: G1 and G2, for which the lower envelopes of the allowable limits of sea states are the operational limits of the combined group of activities. The envelopes also allow to identify the limiting parameters governing the execution of the marine operation. By following the aforementioned procedure, it is possible to systematically establish the allowable limits of the sea states for each group of continuous activities in the entire marine operation. Allowable limits of sea states Activity 1 Hs Activity 1 Activity 2 G1 CONTINUOUS ACTIVITIES Activity 2 Activity 3 Governing activities: 2 & 3 envelope 1 Activity 3 Tp NON-CONTINUOUS ACTIVITIES: 3 & 4 Hs Activity 4 Activity 5 G2 CONTINUOUS ACTIVITIES Activity 4 Activity 5 Governing activities: 5 envelope 2 Tp Figure 2.3: Allowable limits of sea states for groups of continuous activities (Guachamin Acero et al., 2016d) 2.4 Weather window analysis and operability Weather windows are necessary for assessment of the operability of a marine operation during the planning phase, and to make decisions on whether to

51 2.5. Assessment of characteristic values of limiting parameters 23 start or to stop the execution of an activity. For these purposes, hindcast and forecasted wave data are required. For the planning phase, the weather windows can be identified and used to assess the operability of a marine operation. The time histories of Hs and T p based on hindcast wave data are required, see Fig. 2.4 (a). For every time step, the corresponding T p i is used to identify the allowable Hs i for every group of activities, see Fig. 2.4 (b). By comparing the time histories of hindcast Hs and their allowable limits (for corresponding T p), the workable weather windows of each group of activities can be identified, see Fig. 2.4 (c). Then, the weather windows of each activity group are put in sequence and including their respective duration. An example for two groups is shown in Fig. 2.4 (e), where t R1 and t R2 are their duration. The parameter t R is denoted as reference period by Det Norske Veritas (2011a). A starting time for activity group G1 is first identified. After G1 is finished, G2 starts. Since G1 and G2 are not continuous, they can be split. Following this procedure, the workable weather windows of the complete operation can be identified, and by counting them during the total period of analysis, the operability can be assessed. For the execution phase, the weather windows can be identified by following the same procedure as for the planning phase; however, the forecasted wave data need to be used, see Fig. 2.4 (a). The procedure outlined above is ideal and systematic, but does not account for the fact that the starting and stopping times for the activities are uncertain, and thus, it can be conservative or unconservative. However, this topic is not addressed in this thesis. 2.5 Assessment of characteristic values of limiting parameters It was stated earlier that a characteristic value of a limiting (response) parameter needs to be calculated for a target percentile or a non-exceedance probability of an extreme value distribution. This distribution needs to be fitted using the extreme responses, which are calculated by numerically modeling the actual marine operations. The number of seeds for the TD simulations should be sufficient to achieve convergence of the statistical properties (response statistics), and thus, reduce the statistical uncertainty. However, this is only one source of uncertainty and several other sources have to be considered as discussed below. Other sources of uncertainties include numerical models (model uncertainty), wave models (wave model uncertainty), forecasted wave data (fore-

52 Methodology for assessment of operational limits and operability 24 of marine operations 3.3 Figure 2.4: Weather windows analysis including continuity and duration of offshore activities. (a) Hindcast or forecasted Hs and Tp; (b) Allowable limits of Hs for corresponding Tp; (c) Hindcast or forecasted and allowable limits of Hs; (d) Dynamic responses based on forecasted wave data and allowable limits of responses for the monitoring phase; (e) Workable weather windows (Guachamin Acero et al., 2016d)

53 2.5. Assessment of characteristic values of limiting parameters 25 cast uncertainty), human decisions, etc. In all these uncertainty sources, there are modeling parameters for which the probability density functions (PDFs) can be established, for instance, the forecasted Hs. The influence factors of each of these uncertainty modeling parameters on characteristic values of the limiting (response) parameters can be assessed from probabilistic models, and the most important ones can be considered. Then, the characteristic value of a limiting parameter can be computed for a target non-exceedance probability. The target non-exceedance probability should be chosen based on the consequences of failure events and reversibility of the operation. For instance, the non-exceedance probability of the allowable tension of a wire rope during a lifting operation should be low enough to guarantee safety, because consequences of a structural failure can be catastrophic. In contrast, the mating attempt of a transition piece with a monopile foundation should be designed for a higher exceedance probability because this operation can be tried again in the case of a failed attempt (reversible operation) and the consequences are not catastrophic. Thus, the target non-exceedance probability depends on the type of operation. Moreover, it is noted that the responses need to be assessed by modeling the actual operations, so the duration of simulations should be applied according to the operational manuals. Some offshore standards and papers dealing with sources of uncertainty and target non-exceedance probability are briefly discussed below. Det Norske Veritas (2014b) recommends that characteristic values of loads for some marine operations such as loadout, transport, and installation of subsea objects, correspond to a 10 % probability of exceeding the allowable limits (design loads) of limiting parameters. In addition, Det Norske Veritas (2014a) provides load factors for heavy lifting operations. These factors account for failure event consequences and uncertainties in several material and geometric parameters that affect the load and the capacity of the structural components. The intention of these factors is to ensure a probability for structural failure less than per operation. Furthermore, statistical models given by Natskår et al. (2015) or alpha factors provided by Det Norske Veritas (2011a) can be used to account for uncertainties in forecasted Hs as function of forecast lead times. The alpha factors are reduction factors that can be applied on the Hs design limits (operational limits). In fact, Det Norske Veritas (2011a) states that the alpha factors should be calibrated to ensure that the probability of exceeding the operational limits (in terms of Hs) with more than 50% is less than Based on this statement, it is demonstrated that these factors are

54 Methodology for assessment of operational limits and operability 26 of marine operations considered to be independent of the type of operation and consequences of failure events. Moreover, T p needs to be included because it is an important parameter for floating vessels. Thus, distribution functions that account for uncertainties in both Hs and T p parameters are required. In this thesis, the probabilistic assessment of the operational limits including all sources of uncertainty is not addressed. The effects of the statistical uncertainty and consequences of failure events are in some extent included when assessing the characteristic values and allowable limits of limiting (response) parameters, i.e., assuming percentiles for extreme value distributions and safety factors for the capacity.

55 Chapter 3 Installation systems and procedures 3.1 General The installation of foundations, transition pieces and turbine tower components is normally executed with jack-up and floating crane vessels. Jack-up vessels offer a stable platform for crane lifting operations, so the lifting operations do not depend on wave conditions. As it was mentioned earlier, some disadvantages of these platforms are the water depth limitations, mobilization time, and low sea states in which the legs can be deployed and retrieved. Alternatively, floating HLVs are known to be competitive for marginal wind farms and offer fast transit speed between wind turbines. A downside of floating vessels is that the dynamic responses are sensitive to wave actions, so the installation of OWT components is more challenging. The installation activities studied in this thesis are executed using floating HLVs. This chapter deals with the system components and procedures for the MP initial hammering process, TP mating operation, and the entire installation of the tower and RNA. The procedures for the MP and TP installation are commonly applied by offshore installation contractors, while a novel procedure developed by Guachamin Acero et al. (2016c) (paper 5 ) is used to assess the installation of the OWT and RNA. The installation procedures are required for identification of potential critical activities, which is carried out in Ch

56 28 Installation systems and procedures 3.2 Monopile initial hammering process A MP is normally installed using a jack-up or a floating HLV. Monopile foundations are normally transported on the deck of crane vessels, from which they are lifted-off, relocated, upended, lowered and driven into the soil. A potential critical operation for MP installation is the initial phase of the MP hammering process, which starts when a hydraulic hammer is put on the top of the MP and finishes when the MP is able to stand on its own under wave actions. The system components and installation procedure are given by Li et al. (2016b) (paper 2 ) and are summarized below System components The main system components for the MP initial hammering process are shown in Fig A HLV is rigidly connected to a gripper device. The gripper is composed of hydraulic cylinders and piston rods that are radially distributed, see Fig. 4.2 (a). The gripper holds the MP in the vertical position during the initial part of the hammering process. hoses hydraulic hammer HLV gripper MP Figure 3.1: System components for the MP initial hammering process The main particulars of the structures are given in Table 3.1. A HLV with an eight point mooring line system and a MP structure are the main structural components.

57 3.2. Monopile initial hammering process 29 Table 3.1: Main particulars of the structures for the MP initial hammering process (Li et al., 2016b) Parameter Notation Value Units - HLV Displacement tons Length L 183 m Breadth B 47 m Draught T 10.2 m - Monopile Mass M MP 500 tons Diameter D MP 5.7 m Thickness th T P 60 mm - Hydraulic hammer Mass M HAM 300 tons Installation procedure for MP initial hammering process A summary of the installation procedure is given below. 1. The gripper is used to hold the MP after its landing on the seabed. The gaps between the hydraulic pistons rods and the MP are closed to avoid impact loads. The MP and the gripper move in the horizontal plane due to first and second order wave forces. 2. A hydraulic hammer is placed on top of the monopile using the HLV crane. 3. The mooring lines of the HLV are adjusted to position the MP vertically. 4. The mean inclination of the MP is measured at various quadrants of the monopile using a handhold inclinometer. 5. The MP inclination is corrected to a zero mean value by varying the stroke of the piston rods of the hydraulic cylinders. A pre-compression force is applied. 6. To drive the MP into the soil, a few blows are applied using the hydraulic hammer. The MP inclination, which is caused by wave actions, is measured and corrected by varying the stroke of the piston rods. 7. The previous step is repeated until the hydraulic cylinders are not able to correct the MP inclination.

58 30 Installation systems and procedures 8. Correct the MP inclination by varying the length of the mooring lines and by applying thruster forces. 9. Retract the hydraulic piston rods and drive the MP to final penetration. 3.3 TP mating operation using a floating crane vessel The TP is a cylindrical structure that is used to connect the MP foundation with the tower lower section. The TP is normally transported on the deck of an installation vessel. Once the MP is installed, the TP will be lifted off, aligned with the MP and lowered until it lands on the MP foundation. A critical activity is normally the mating operation. It starts when the TP bottom tip is located a couple of meters above the MP, and ends when the TP bracket supports land on the MP tip, see Fig Guachamin Acero et al. (2016b) (paper 4 ) studied the mating operation between a TP structure and a MP foundation. The main particulars of the system components and the installation procedure are summarized below System components Figure 3.2 shows that the system is composed of a HLV, a TP structure with a rigging system connected to the crane tip, and a MP foundation fixed to the seabed. The main particulars of the HLV and TP are shown in Table 3.2. The MP was modeled according to Table Installation procedure for the TP mating operation The installation procedure for the TP mating operation is summarized as follows: 1. The TP structure is aligned with the MP structure and positioned a couple of meters above the MP. 2. The motions of the TP bottom tip are monitored to decide whether or not the mating operation is possible. 3. If the motions of the TP bottom tip are acceptable, the TP is slowly lowered towards the MP.

59 3.4. Novel concept for OWT tower and RNA installation 31 hoist wire spreader bar slings slings hook TP spreader bar HLV MP Figure 3.2: System components for the TP mating operation Table 3.2: Main particulars of the structures for the TP mating operation (Guachamin Acero et al., 2016b) Parameter Notation Value Units - HLV Displacement tons Length L 140 m Breadth B 30 m Draught T 6 m - Transition piece Mass M T P 200 tons Diameter D T P 6.4 m Thickness th T P 60 mm Length L T P 20 m 4. The initial part of the mating operation occurs. 5. The lowering process continues until the bracket supports of the TP land on the MP tip. 3.4 Novel concept for OWT tower and RNA installation Guachamin Acero et al. (2016c) (paper 5 ) developed a novel concept and procedure for installation of the fully assembled OWT tower and RNA.

60 32 Installation systems and procedures This concept is based on the inverted pendulum principle and was shown to be a feasible alternative for jack-up crane vessels. Moreover, this concept eliminates the necessity of huge floating crane vessels in terms of lifting capacity and height. Typical OWT foundations can be monopiles, tripods and jackets (under specific modifications) System components Figure 3.3 shows the main components of the installation system. A medium size HLV is required for upending of the OWT. In addition, a cargo barge is needed for transportation of the OWT to the site in an horizontal position, which is assumed to be feasible. Moreover, a specially designed upending frame is also required to upend the tower. This upending frame has two grippers. The grippers provide a pair of supports that are able to transfer the bending moments and shear forces. The foundations need to be installed with an specific heading based on wave spectra statistics. This is because of the benefits of shielding effects provided by the HLV on the motions of the cargo barge. gripper OWT tower and RNA upending frame upending frame bottom pin hinged support HLV hoist wire cargo barge tripod hinged saddle support Figure 3.3: System components during lift-off of the OWT tower and RNA The main particulars of the system components are given in Table Installation procedure A summary of the procedure for installation of the OWT tower and RNA is given below.

61 3.4. Novel concept for OWT tower and RNA installation 33 Figure 3.4: Installation procedure for OWT tower and RNA. (a,b) Motion monitoring phase; (c) Mating between the upending frame and the foundation supports; (d) Upending of the OWT tower in the final stage; (e) Completion and upending frame removal

62 34 Installation systems and procedures Table 3.3: Main particulars of the structures for OWT and RNA installation (Guachamin Acero et al., 2016c) Parameter Notation Value Units - HLV (see Table 3.2) - Cargo barge Displacement tons Length L 69 m Breadth B 18 m Draught T 4 m Metacentric height GM 3.75 m Vertical position of COG above keel V CG 5.0 m - Tower components (Jonkman et al., 2009) Tower mass M tower 348 tons Nacelle mass M nacelle 240 tons Blades mass M blades 110 tons 1. The stern of the cargo barge is moored to the foundation using mooring lines and fenders. The bow of the barge is moored to the seabed using typical catenary lines, see Fig. 3.4 (a). 2. A HLV is moored to the seabed using a pipelaying mooring arrangement. This is because the vessel has to move forward during the OWT installation. 3. The separation between the stern of the barge and the foundation is adjusted using cantilevered beams. This allows for alignment between the upending frame bottom pins and the supports on the foundation (docking cones). 4. The motions of the upending frame bottom pins are monitored to decide whether or not to start the operations. 5. The crane tip or the lifting block and the lifting point on the OWT tower are aligned and connected with lifting wires in slack condition, see Fig. 3.4 (b). 6. The lift-off operation starts and continues until the bow saddle support on the cargo barge is cleared, see Fig The mating process between the bottom pins of the upending frame and the foundation supports starts. This operation is finished when

63 3.4. Novel concept for OWT tower and RNA installation 35 there is enough clearance between the OWT and the stern saddle support on the cargo barge, see Fig. 3.4 (c). 8. The lifting points are aligned by gradually moving the crane vessel forward. This is done by pulling-in mooring lines connected to the seabed. This is a well-known procedure employed by pipelay vessels in shallow water. 9. The OWT upending process continues by moving the HLV in small increments. 10. By the final stage of the upending phase, the lifting points do not need to be aligned, because an horizontal force is required to bring the OWT to vertical position and avoid a possible crane interference, see Fig. 3.4 (d). 11. The OWT is lowered using hydraulic jacks provided in the upending frame, the turbine tower lands on the foundation, where it is finally bolted. 12. The upending frame is removed and placed onto the cargo barge. The frame can be used for another OWT tower and RNA, see Fig. 3.4 (e). Guachamin Acero et al. (2016c) showed that the total installation time can be less than 16 hours, which could be reduced to 8 hours if a DP vessel is employed, because the mooring activity for the HLV would not be required. In contrast, the installation time using jack-up vessels is approximately 3 days (Herman, 2002). Moreover, jack-up vessels can normally operate between 30 to 60 m water depths (El-Reedy, 2012). Thus, this novel concept can offer a higher installation rate and is less water depth-sensitive than the current installation practice.

64 36 Installation systems and procedures

65 Chapter 4 Numerical modeling of offshore wind turbine installation activities 4.1 General The installation procedures for MP, TP and tower and RNA were summarized in Ch. 3. The operational limits of potential critical installation activities will be addressed in Ch. 5. These limits can be established based on quantitative assessment of the dynamic responses. To study these responses, numerical methods and models of the installation systems are required. This chapter deals with numerical methods and models, which are required to study dynamic responses of potential critical OWT installation activities. 4.2 Numerical methods for analysis of marine operations It was shown in Subsec that the allowable limits of sea states are established based on the allowable limits and characteristic values of the limiting (response) parameters. The characteristic values are obtained by numerically modeling the various marine operations. This is done by applying numerical methods and models. In general, frequency and time domain methods are employed; however, more accurate and efficient operationspecific numerical analysis methodologies can be applied. This section deals with numerical methods that were employed for analysis of OWT turbine 37

66 38 Numerical modeling of offshore wind turbine installation activities installation activities Time and frequency domain methods For a dynamic coupled model of an installation system, the equations of motion can be solved in the frequency or time domain. The frequency domain method can be applied to weakly non-linear systems and for analysis of responses resulting from a stationary process, see e.g. dynamic system 1 in Fig In contrast, if the dynamic systems are non-linear and the resulting processes are non-stationary, time domain methods need to be applied. There are various computer codes for analysis of marine operations. In this thesis, the state-of-the-art programs ANSYS-AQWA and SIMO developed by Century Dynamics-Ansys Inc. (2011) and MARINTEK (2012) were employed. The codes are widely used in the offshore industry and academia, see e.g., Oosterlaak (2011), and are capable of representing multi-body dynamic systems, hydrodynamic interaction between diffracting structures, joints with different types of constraints, mooring lines and winches for lifting operations, and fenders. Moreover, external forces can be added into the time domain solver by using an external dynamic library which is suitable for modeling DP and steering systems. Based on these features, the codes are suitable for modeling OWT installation activities. The frequency domain method In the frequency domain, the equations of motion of a rigid body dynamic system can be written as follows: ω 2 [M + A (ω)] X (ω) + iωb (ω) X (ω) + KX (ω) = F ext (ω) (4.1) Where, M is the multi-body structural mass matrix, A and B are the frequency dependent added mass and damping matrices. The coefficients of these matrices are normally calculated by diffraction analysis using the panel method. K is the stiffness matrix, which accounts for the hydrostatic, mooring and coupling restoring terms. X is the vector of the motion of the system and F ext is the vector of external forces acting on the system. As stated earlier, in this thesis only the first and second order wave actions are considered. The equations of motions can be solved for a response parameter under regular wave actions. The result is a transfer function or response amplitude operator (RAO) which can be used to carry out spectral analyses and assess the system dynamic responses in irregular seas.

67 4.2. Numerical methods for analysis of marine operations 39 The time domain method In the time domain, the equations of motion of a dynamic system can be written as follows: [M + A ] ẍ (t) + t 0 h (t τ) ẋ (τ) dτ + Kx (t) = F ext (x, ẋ, t) (4.2) Where A is the infinite frequency added mass matrix, x, ẋ, ẍ are the position, velocity and acceleration of the system, and the added mass and damping coefficients are incorporated in the equations of motion via the retardation function h. The various coupling forces due to lifting wires and fender elements are included in the stiffness matrix. The equations of motion are solved in every time step increment by applying numerical techniques, e.g., Newmark-beta methods Dynamic systems and numerical methods for analysis of OWT installation activities Marine operations need to be analyzed by modeling numerically the actual operations. In addition, a dynamic system can be non-linear and the response can be a result of a non-stationary process. Furthermore, since marine operations require moving objects from one place to another, changes such as winch speed need to be imposed into a dynamic system. As a consequence of these changes, the dynamic properties of a system may change. This subsection addresses typical dynamic systems and standard numerical methods required for analysis of OWT installation activities. Weakly non-linear and non-linear dynamic systems Some phases of an offshore operation correspond to weakly non-linear systems. A typical example is the monitoring phase prior to execution of a lifting operation, see the dynamic system (a) in Table 4.1. Another example is the motion monitoring phase of a vessel prior to a mating operation, see Subsec These installation phases can be solved in the frequency domain, because the motions are small, the dynamic properties of the system do not change in time, and the system is weakly non-linear. There are also non-linear systems in which the output does not vary linearly with the input. For these systems, the stochastic responses need to be computed by applying the time domain method. The dynamic systems (b,c,e) in Table 4.1 are typical examples of OWT installation activities where the time domain method is normally applied.

68 40 Numerical modeling of offshore wind turbine installation activities Table 4.1: Dynamic systems, stochastic dynamic responses and numerical methods for analysis of typical OWT installation activities. Mean value (µ) and standard deviation (σ) are statistical properties of dynamic responses Time-invariant (a) Pre-lift Dynamic system Stochastic dynamic responses Numerical method z [m] Crane tip motion Frequency or time domain t [s] Stationary process: dσ/dt = 0; dµ/dt = Wire tension Time domain Time-variant F hoist [N] t [s] non-linear Non-stationary process: varying module position (c) Mating Impact force Time domain F imp [N] t [s] non-linear Non-stationary process: varying module position (d) Lowering 62 Lifting block position Time domain z block [m] weakly nonlinear (b) Lift-off weakly nonlinear (e) MPhammering t [s] Non-stationary process: dσ/dt 0; dµ/dt 0,varying block position Gripper-MP contact force Time domain F [N] t [s] non-linear Non-stationary process: varying MP penetration

69 4.2. Numerical methods for analysis of marine operations 41 Stationary and non-stationary processes The wave elevation is said to be a stationary process if the statistical properties such as the mean value and standard deviation of a realization of that process (e.g., registration of the wave elevation) are independent of time. These statistical parameters are used to represent the sea state by means of a wave spectrum. For analyses of marine operations, this wave spectrum is normally assumed to be invariant for a period of 3 hours. During offshore lifting operations, the dynamic system always experiences changes imposed by an external action. For example, a winch is required to slew the crane, to lower a payload or to move the vessel to a new position. By taking the installation of a topside module as an example, an imposed winch speed makes the lowering process non-stationary. As a result, the statistical properties of the position of the topside module vary with time. Thus, imposed changes on a system under stationary wave conditions can make the output of the system a non-stationary process, see the responses of the dynamic system (d) in Table 4.1. If the imposed changes significantly affect the dynamic properties of the system (e.g., natural periods), the system is time-variant. In general, offshore lifting and lowering operations need to be solved in the time domain even if the system is weakly non-linear. However, if the the winch speed is applied very slowly, so that the dynamic properties of the system do not change significantly and the system is weakly non-linear, the operations can be studied using stationary process TD simulations (condition when the system has reached the steady state). Imposed changes such as winch speed, can be added separately to the time histories of the dynamic response. This method is addressed below Methodologies for analysis of lift-off and mating operations As it was stated earlier, it is important to develop efficient methodologies that allow a quick assessment of the dynamic responses. In this subsection, two methodologies: one for lift-off and landing, and another one for mating operations are addressed. These methodologies are general and relevant for OWT installation activities Lift-off and landing operations using floating heavy lift vessels In offshore installation, most of the dynamic systems have time-variant dynamic properties. As stated earlier, external actions (e.g., winch speed)

70 42 Numerical modeling of offshore wind turbine installation activities change the state of a dynamic response. In these cases, the system dynamic properties are greatly influenced by these external actions, so the dynamic responses need to be assessed by simulating the actual operations. In contrast, there are systems in which these changes happen so slowly that they can be analyzed by applying stationary process TD simulations. Systems with time-invariant dynamic properties The numerical methods given here are suitable for dynamic systems on which a change has been imposed (e.g. winch speed) on the dynamic response computed from a system in steady state condition, and the dynamic properties of the system are not significantly affected. It has been shown that limiting (response) parameters such as snap and landing velocities are suitable for characterizing lift-off and landing operations (Det Norske Veritas, 2011b; Guachamin Acero et al., 2016b,c). This is because these parameters are directly related to the snap and impact forces. A suitable numerical method (collision method) for assessment of the first snap or impact velocities was proposed by Sandvik (2012). This method was originally developed by Fylling (1994) where it was applied to estimate collision velocities of drifting ships. Guachamin Acero et al. (2016b) (paper 4 ) applied the method for assessment of the landing and wire snap velocities during a TP installation onto a MP foundation. Figure 4.1 shows the basis for estimating impact velocities. It requires TD histories of the relative displacement z rel and velocity v rel of the mating points (from coupled models and stationary process TD simulations). The lowering speed is imposed into the relative displacement time history, and only points having a negative relative displacement (z imp ) are considered, see Fig. 4.1 (a) red color points. The corresponding velocities v imp without winch speed are shown in Fig. 4.1 (b). Note that snap loads can occur after landing of the structure if the crane tip is moving upwards and the slack length of the wire is short. This is possible when the distance paid-out by the crane wire is smaller than the displacement of the crane tip. Figure 4.1 (a) shows the points z snap complying with this condition (black color). The snap velocities v snap (without including winch speed) are shown in Fig. 4.1 (b). For lift-off operations, the same reasoning is applied. The winch has to pay-in and the crane tip has to move upwards. The total snap velocity is the sum of a characteristic value of the dynamic snap velocity and the winch speed. Because lift-off and landing activities last only a few minutes, the time history of the velocity can be divided into several short duration intervals. Then, extreme value response statistics and distributions can be established.

71 4.2. Numerical methods for analysis of marine operations 43 The number of samples should ensure that statistical properties such as mean value and standard deviation have converged. Based on the extreme value response statistics or distributions, a characteristic (response) value corresponding to a target probability of non-exceedance or percentile is selected. This target probability depends on the type of operation and associated failure events, see Sec This numerical method has been applied to identify heave impact and snap velocities during the TP installation, see Subsec z [m] z rel excluded points z imp -47 z snap reference level for next reference level for next impact velocities snap velocities a) Relative displacements for impact and snap events 0.4 v rel v [m/s] v imp v snap t [s] b) Relative impact and snap velocities Figure 4.1: Determination of impact and snap velocities based on TP bottom tip displacement and velocity TD histories. (a) TP bottom tip relative heave displacements (respect to the crane tip) for constant lowering speed; (b) Possible impact and snap velocities (no winch speed included) (Guachamin Acero et al., 2016b) Systems with time-variant dynamic properties During installation of OWT components, the MP, TP and tower structures may not be considered to be light weight, because the installation vessels are normally smaller than the ones used for the oil and gas industry and their responses can be affected. If this is the case, lift-off operations need to be analyzed by imposing the winch speed in the equations of motion and solving them in the time domain. In other words, it is necessary to model the actual operations. Typical examples are the lift-off of a module from a cargo barge and the mating of heavy topside modules. In these activities, the dynamic properties of the structures are different before, during and after

72 44 Numerical modeling of offshore wind turbine installation activities the load transfer. In addition to the non-linear system, changes in stability parameters may occur, and these effects need to be included. Thus, several TD simulations of the actual operations using random seeds are required for assessing extreme value distributions, and thus, estimating characteristic values, see Sec Mating operations Mating operations are standard activities within the oil and gas, and wind energy industries. Mating requires a docking pin to be inserted into a female receptacle or docking cone. The available mating gap is given by the annular space between the mating structures. The current practice is to calculate the significant or 3-hour maximum responses by applying stationary process TD simulations. Then, these response parameters are compared with the allowable mating gap to decide whether a sea state is acceptable or not. However, the actual monitoring phase prior to mating operations only lasts a few minutes and the surge and sway motions are correlated. A practical criterion to decide whether or not a mating operation is possible was proposed by Guachamin Acero et al. (2015) (paper 3 ). This method is based on the number of crossings that a mating pin is allowed to cross circular boundary of radius r in a time interval. This criterion is practical and has been adopted in this thesis. Furthermore, the magnitude of the mating gaps and allowable crossing rates depend on how well the structures are aligned and the criterion of the operators. These are sources of uncertainties that are not addressed in this thesis, but are discussed in Sec The numerical solution for the crossing rate is obtained from spectral analyses of response parameters that are computed by applying the frequency domain method. An alternative solution has also been proposed by Low (2009) using another approach, and it was applied to a different problem. It was found that both 1 st and 2 nd order motion contributions of the TP should be included. Both contributions were assumed to be independent, so their covariance matrices can be added. By considering that the allowable crossing rates required for mating operations are relatively high (1 or 2 crossings per minute), the Gaussian approximation for the second order wave motions will provide acceptable results when compared with a more sophisticated method such as the time domain. The numerical solution is fast and accurate, which makes it suitable for quick assessment of the allowable sea states. Out-crossing rate from a circular boundary The limit state function defined by a circle could be written in the following

73 4.3. Numerical modeling of the MP initial hammering process 45 form: ( x1 ) 2 ( x2 ) 2 D(X) = + = 1 (4.3) r r Where, x 1 and x 2 are the stochastic responses in surge and sway. Naess (1983) proposed a general methodology to evaluate the out-crossing rate of a vector process based on the properties of the conditional probability. This methodology was applied to equation (4.3) and the following solution was obtained (Guachamin Acero et al., 2015). ν + S D = E (ẋ n X = x) f X (x 1 = f(x 2, d), x 2 ) J dx 2 (4.4) x 2 [ E (ẋ n X = x) = ẋ n fẋn dẋ n = σẋn exp 1 ( ) ] 2 ( ) mẋn mẋn + mẋn Φ 2 σẋn σẋn 0 J = (4.5) 2r [d (x 2 /r) 2] 0.5 (4.6) Expression (4.4) can be evaluated numerically, ν + S D is the outcrossing rate, J is the Jacobian of the transformation of x 1 in the bivariate probability density function (PDF), which needs to be evaluated for d = 1 following the definition of the limit state function, r is the radius of the circle, Φ is the Gaussian cumulative distribution function (CDF) and σẋn & mẋn are the standard deviation and mean values of the normal velocities when solved for their conditional PDF. Alternatively, TD simulations of the stationary process can be applied. The numerical solution was applied for assessment of the allowable limits of sea states during the TP and OWT tower and RNA mating activities, see Subsecs and Numerical modeling of the MP initial hammering process The (actual) MP driving and correction of the inclination require modeling of the following parameters: the hammer blow impact force, the hydrodynamic forces acting on the MP and HLV, the gripper contact forces, the soil-mp reaction forces and the correction forces. As stated in the installation procedure, the MP-gripper system moves in the horizontal plane due to the first and second order wave actions. The problem is simplified by assuming that the hammer blows do not affect the inclination of the MP and

74 46 Numerical modeling of offshore wind turbine installation activities by decomposing the total gripper contact force into a dynamic part resulting from wave actions acting on the HLV-MP system, and MP inclination correction force required to bring the MP to a zero mean inclination. The dynamic component is obtained from global dynamic analysis while the correction force is calculated from quasi-static analyses (Li et al., 2016b) (paper 2 ). The dynamic coupled models were built in the SIMO program (MAR- INTEK, 2012). The dynamic responses including the gripper contact force were studied by conducting stationary process TD simulations at 2 m stepwise incremental penetrations between 2 and 10 m. The coupled equation of motions were solved in the time domain for incremental steps of s. The added mass and damping coefficients were calculated from WAMIT using the panel method and incorporated in the time domain via retardation functions. The external actions on the system are the first and second order wave forces. Coupling and mooring forces were included in the stiffness and damping matrices. The hydraulic system of the gripper device was modeled using spring and damper elements. The spring coefficient for the MP-gripper contact elements k grip = N/m was estimated based on guidelines given by Albers (2010) for typical hydraulic cylinder specifications (IHC Merwede, 2016). The damping coefficient was chosen as b grip = 0.2b cr. The elastic contact model is shown in Fig. 4.2 (a). The soil-mp interaction of large diameter piles at shallow penetration depths is different from slender piles at large penetration depths. For the former one, it has been shown that the skin friction and base shear provide large contributions to the overall soil reaction forces (Byrne et al., 2015; Lesny and Wiemann, 2006). For the dynamic coupled system, the Winkler model was applied; therefore, the soil-mp interaction was modeled with non-linear springs and dampers. The contributions from the lateral load-deflection P-y, friction T-z, base shear F shear, and MP end tip loaddisplacement Q-z curves were applied. These curves were derived based on recommendations given by the American Petroleum Institute (2000) for typical sandy soils. The spring-damper models are shown in Fig. 4.2 (b). Since the soil-mp skin friction was included, some elastic energy will be dissipated during the loading and unloading phases. An example of an hysteretic loop for MP inclination under cyclic loading is shown in Fig The hammer was modeled as a point mass on top of the monopile. For the correction phase, quasi-static analyses of the correction force were conducted. This force is due to the intact soil reaction when moving the MP back to a zero mean inclination, and was calculated for the expected

75 4.3. Numerical modeling of the MP initial hammering process 47 Figure 4.2: Spring-damper models. (a) Gripper-MP physical and contact models; (b) Soil-MP interaction model (Li et al., 2016b) 3000 Soil reaction momoent [knm] MP inclination [deg] Figure 4.3: Typical soil reaction moment vs MP inclination due to cyclic loading with a period of 6 s (Li et al., 2016b) maximum MP inclination (obtained from global response dynamic analysis) during a 10 min duration of the operation, which corresponds to the elapsed

76 48 Numerical modeling of offshore wind turbine installation activities time between a correction phase and a hammering event. outline of a dynamic coupled model is shown in Fig A schematic Figure 4.4: Schematic outline of a dynamic coupled model for the MP initial hammering process (Guachamin Acero et al., 2016d) 4.4 Numerical modeling of the TP mating operation The TP mating operation is divided into several activities, which were modeled numerically according to the methodology proposed by Guachamin Acero et al. (2016b) (paper 4 ). A summary of the methodologies for numerical analysis of potential critical TP installation activities, numerical methods and modeling parameters that were used to model the various activities is given in Table 4.2. The dynamic coupled models were built in the AQWA suite of computer codes developed by Century Dynamics-Ansys Inc. (2011). A schematic outline of a dynamic coupled model is shown in Fig For analyses of TP installation activities using the TD method, a time step of s was applied. The potential critical installation activities provided in Table 4.2 are identified in Subsec A brief description of the numerical methods and elastic contact models used to model these activities is provided below. Activity 1: The TP bottom tip motion monitoring phase is studied by applying the crossing rate method given in Subsec This is possible

77 4.4. Numerical modeling of the TP mating operation 49 Table 4.2: Numerical methods and modeling parameters for the TP mating activities (Guachamin Acero et al., 2016b) No. Activity Methodology Numerical method Modeling parameters 1 Monitor TP s bottom motions prior to mating Crossing method, rate Subsec. 2 Mating operation Collision method, Subsec FD annular gap r = 0.30 m (Assumed) TD winch lower. speed = 2 m/min, duration 1 min 3 Lowering N. A. TD winch lower. speed = 2m/min, duration 5 min 4 Landing Collision method, Subsec TD winch lower. speed = 2m/min, duration 1 min because the motion responses of the TP bottom tip are small and they are the result of a stationary process. Activity 2: During the initial phase of the mating operation, axial and lateral impact of the finger guides can occur. The axial impact is studied by applying the collision method described in subsection Note that in this method, the impact events are not modeled, and thus, only characteristic values of impact velocities can be assessed. The physical and contact models as well as the possible contact-impact points are shown in Fig The lateral impact was studied by applying stationary process TD simulations. The contact model shown in Fig. 4.7 was applied. Activity 3: For analysis of the TP lowering phase, non-stationary process TD simulation were carried out. The elastic contact model shown in Fig. 4.7 was applied as well. Activity 4: For the landing phase, the collision method given in subsection was applied. For comparison, non-stationary process TD simulations were conducted, see Subsec The physical and contact models are shown in Figure 4.8. The rigging configuration was chosen according to Fig. 4.6 (a). The elastic wires and contact elements were modeled according to Table 4.3. Contact elements were required for non-linear global dynamic analyses of installation activities. The stiffness coefficient for the contact elements k con was reasonably estimated based on the pile head load-displacement curves given by Hoit et al. (2007) and the T-z curves provided by the American Petroleum Institute (2000). For a pile with a diameter of 6 m and a penetration depth of 40 m in soft clay, the secant axial stiffness can be inferred

78 50 Numerical modeling of offshore wind turbine installation activities Figure 4.5: Schematic outline of a dynamic coupled model for TP landing (Guachamin Acero et al., 2016b) to be approximately kn/m for load levels of 6000 kn. For sandy soils it can be between 1 and kn/m. Table 4.3: Summary of spring and damper coefficients for numerical analyses of the TP installation activities (Guachamin Acero et al., 2016b) Parameter Notation Value Units Properties Source Hoist wire rope stiffness Wire rope sling stiffness Contact elem. spring coeff. Contact elem. damping coeff. TP surge, sway damping coeff. k wire kn/m D=60mm, L=40m, N=4, MBL=3.5MN, E=210 GPa k sling kn/m D=60mm, L=10m, N=1, MBL=3.5MN, E=210 GPa k wire = N EA/L, (Lankhorst ropes, 2015) k con kn/m (American Petroleum Institute, 2000; Hoit et al., 2007) b con 0.05b cr - Assumed b 11, b cr - Assumed 4.5 Numerical modeling of the OWT tower and RNA installation activities Guachamin Acero et al. (2016c) (paper 5 ) provided modeling parameters

79 4.5. Numerical modeling of the OWT tower and RNA installation activities 51 Figure 4.6: Models for TP-MP axial contact-impact prior to mating. (a) Physical model; (b) Contact points; (c) Spring-damper model (for Activity 2) (Guachamin Acero et al., 2016b) Figure 4.7: Lateral contact during TP mating operation (a) Physical model; (b) Spring-damper model (for Activities 2, 3) (Guachamin Acero et al., 2016b) for the various installation activities. The entire installation procedure was

80 52 Numerical modeling of offshore wind turbine installation activities Figure 4.8: Contact-impact during TP landing (a) Physical model; (b) Spring-damper model (for Activity 4) (Guachamin Acero et al., 2016b) divided into several potential critical activities, see Subsec This is normally done to facilitate the modeling of complex marine operations. The OWT tower and RNA installation activities are analyzed in the AQWA suite of computer codes. A summary of these activities, methodologies, numerical methods and some numerical modeling parameters are given in Table 4.4 and described in more detail below. A schematic outline of the dynamic coupled model during frame mating is shown in Fig Activity 1: Prior to the lift-off operation, the total weight of the tower and the upending frame is taken by two saddle supports on the cargo barge, see Fig. 3.4 (b). These supports were modeled using springs and dampers to restrict vertical and lateral motions and rotations with respect to the cargo barge, see Fig (a). For the OWT tower lifting operation, a winch speed of 5 m/min was assumed. Once the lift-off operation is completed, the winch stops, the saddle support located at the bow of the cargo barge clears the tower, and the hinged saddle support located at the stern of the cargo barge is still loaded. The activity was simulated by applying the time domain method. Activity 2: There exist a phase where the upending frame pins are aligned and the motions are monitored. This process is stationary, so the

81 4.5. Numerical modeling of the OWT tower and RNA installation activities 53 Table 4.4: Activities, numerical methods and modeling parameters for the OWT tower and RNA installation (Guachamin Acero et al., 2016c) No. Activity Methodology Numerical method Modeling parameters 1 OWT tower liftoff 2 Monitor upending frame bottom pin motions 3 Upending frame and foundation support mating N. A. TD hoist winch speed = 5 m/min, duration 1 min Crossing method, rate Subsec. FD docking cone radius r = 0.35 m N. A. TD hoist winch speed = 5 m/min, duration 1 min 4 OWT upending N. A. TD variable HLV forward speed 3 m/min, hoist winch speed = 3 m/min, duration 30 min Figure 4.9: Schematic outline of a dynamic coupled model for the tower liftoff and upending frame mating operations (Guachamin Acero et al., 2016c)